Investigation of NIR spectra pre-processing methods combined with multivariate regression for determination of moisture in powdered industrial egg

Abstract

High productivity and meantime perishability of in natura eggs, make powdered egg attractive for patisseries and pasta industries. Water reduction in 65%, extends shelf life from 1 to 12 months, preventing also Salmonella. Maximum powdered egg moisture allowed by Brazilian law is 6.0% (w w-1). However, its determination by reference technique (oven at 105ºC for 8 hour) is lengthy for processing industry. Therefore, the purpose of this study was to investigate the influence of several spectral pre-processing techniques in the application of near-infrared spectroscopy associated with chemometrics models for determination of moisture content in powdered egg, without the need of sample preparation and destruction, held at 0.5 min. Several pre-treatment techniques were evaluated to ensure spectral data reliability such as: standard normal variation; multiplicative scatter correction; smoothing and detrend. The principal component regression (PCR) and partial least squares (PLS) were evaluated with and without pre-treatment. The best results were observed in NIR/PLS model (49 samples), providing an adequate correlation (r) of 0.96, for cross-validation. Using 21 samples as prediction set, NIR/PLS showed relative error (RE) < 2.0%, compared to primary methods oven and thermobalance, indicating to be suitable for industrial quality control.